Effect of Cytoplasmic Domain Mutations on the Agonist-stimulated Ligand Binding Activity of the Platelet Integrin αIIbβ3

Function of the platelet integrin αIIbβ3 is regulated by agonist-generated signals interacting with its cytoplasmic tails. When αIIbβ3 is expressed in Epstein-Barr virus-transformed B lymphocytes, stimulation of the cells with phorbol 12-myristate 13-acetate results in αIIbβ3-mediated lymphocyte adherence to immobilized fibrinogen, as well as soluble fibrinogen binding to αIIbβ3, indicating that agonists increase the affinity of αIIbβ3 for fibrinogen in these cells. To address the contribution of the αIIb and β3 cytoplasmic tails to this process, we mutated each tail and expressed the mutants in B lymphocytes. Truncation of the αIIb tail did not impair unstimulated or stimulated lymphocyte adherence to fibrinogen, regardless whether the truncation was proximal or distal to the conserved GFFKR sequence. However, deleting GFFKR or replacing it with alanines markedly reduced αIIbβ3 expression due to impaired intracellular assembly of αIIbβ3 heterodimers, probably due to a mutation-induced change in the conformation of αIIb. Introducing β3 mutations known to impair αIIbβ3 function in platelets into the cytoplasmic tail of β3 in lymphocytes also impaired αIIbβ3 function in these cells. These studies demonstrate that the cytoplasmic tail of αIIb is not required for αIIbβ3 function in lymphocytes, although the presence of GFFKR in the αIIb tail is required for αIIb to interact with β3. Additionally, they indicate that signals interacting with the β3 cytoplasmic tail are responsible for the ability of agonists to stimulate αIIbβ3 function.

Function of the platelet integrin ␣IIb␤3 is regulated by agonist-generated signals interacting with its cytoplasmic tails. When ␣IIb␤3 is expressed in Epstein-Barr virus-transformed B lymphocytes, stimulation of the cells with phorbol 12-myristate 13-acetate results in ␣IIb␤3-mediated lymphocyte adherence to immobilized fibrinogen, as well as soluble fibrinogen binding to ␣IIb␤3, indicating that agonists increase the affinity of ␣IIb␤3 for fibrinogen in these cells. To address the contribution of the ␣IIb and ␤3 cytoplasmic tails to this process, we mutated each tail and expressed the mutants in B lymphocytes. Truncation of the ␣IIb tail did not impair unstimulated or stimulated lymphocyte adherence to fibrinogen, regardless whether the truncation was proximal or distal to the conserved GFFKR sequence. However, deleting GFFKR or replacing it with alanines markedly reduced ␣IIb␤3 expression due to impaired intracellular assembly of ␣IIb␤3 heterodimers, probably due to a mutation-induced change in the conformation of ␣IIb. Introducing ␤3 mutations known to impair ␣IIb␤3 function in platelets into the cytoplasmic tail of ␤3 in lymphocytes also impaired ␣IIb␤3 function in these cells. These studies demonstrate that the cytoplasmic tail of ␣IIb is not required for ␣IIb␤3 function in lymphocytes, although the presence of GFFKR in the ␣IIb tail is required for ␣IIb to interact with ␤3. Additionally, they indicate that signals interacting with the ␤3 cytoplasmic tail are responsible for the ability of agonists to stimulate ␣IIb␤3 function.
The platelet integrin ␣IIb␤3 (glycoprotein IIb-IIIa, CD41/ CD61) contains a binding site for protein ligands such as fibrinogen and von Willebrand factor that is exposed by platelet stimulation (1). Occupation of the ligand binding site, in turn, results in platelet aggregation, presumably by cross-linking adjacent stimulated platelets. The exposure of the ligand binding site on ␣IIb␤3 results from a metabolic process termed "inside-out" signaling or "activation" in which agonist-generated intraplatelet signals induce a conformational change in ␣IIb␤3 by interacting with its cytoplasmic tails (2). The identity of the signals that activate ␣IIb␤3 and the mechanism by which they alter the conformation of the heterodimer are not known.
The ability of naturally occurring mutations involving the cytoplasmic tail of ␤3 to abrogate ␣IIb␤3 function in platelets (3,4) indicates that this portion of ␤3 is essential in the process of ␣IIb␤3 activation. The role of the ␣IIb cytoplasmic tail is less certain. Although no naturally occurring ␣IIb mutations involving the ␣IIb tail have been reported, studies of recombinant ␣IIb␤3 expressed in Chinese hamster ovary (CHO) 1 cells suggest that the conserved membrane-proximal sequence GFFKR in the ␣IIb cytoplasmic tail regulates the affinity of ␣IIb␤3 for ligands by maintaining unstimulated ␣IIb␤3 in an inactive state (5,6).
To develop an in vitro model to study the agonist-induced function of wild-type and mutant ␣IIb␤3, we have expressed ␣IIb␤3 in human Epstein-Barr virus (EBV)-transformed B lymphocytes and found that stimulation of these cells by phorbol 12-myristate 13-acetate (PMA) resulted in their adherence to immobilized fibrinogen, suggesting that PMA stimulation had increased either the avidity or affinity of ␣IIb␤3 for fibrinogen (7). In this paper, we report that PMA stimulation resulted in the binding of soluble fibrinogen to lymphocytes expressing ␣IIb␤3, indicating that agonist stimulation increased the affinity of the expressed ␣IIb␤3 for fibrinogen. We then used the lymphocyte system to address the roles of the ␣IIb and ␤3 cytoplasmic tails in ␣IIb␤3 activation. We found that although deletion of the ␣IIb tail did not affect the ability of ␣IIb␤3 to interact with fibrinogen or to respond to PMA stimulation, deletion or replacement of GFFKR markedly reduced the expression of ␣IIb␤3 on the lymphocyte surface by impairing the intracellular association of ␣IIb with ␤3. In contrast, introduction of the naturally occurring ␤3 mutations that impair ␣IIb␤3 function in platelets into the cytoplasmic tail of ␤3 expressed in lymphocytes also impaired ␣IIb␤3 function in these cells. These results suggest that intracellular signals interacting with the cytoplasmic tail of ␤3, but not the cytoplasmic tail of ␣IIb, are responsible for the ability of agonists to stimulate ␣IIb␤3 function.

EXPERIMENTAL PROCEDURES
Construction of ␣IIb and ␤3 Mutants-Mutations were introduced into the cytoplasmic tail of ␣IIb and into residue 752 of ␤3 (Fig. 1) using a modified overlap polymerase chain reaction procedure as described previously (7) with oligonucleotides encoding the desired mutations plus additional conservative mutations that generated diagnostic restriction sites (7). Wild-type ␣IIb and ␤3 in the EBV-based episomal plasmids pREP9 and pREP4 (8), respectively, were used as templates for the polymerase chain reaction. The amplified DNA fragments were then subcloned into pBS, sequenced, and assembled into the expression plasmids. The ␤3 mutation Arg-724 3 stop was produced using the QuikChange site-directed mutagenesis method (Stratagene) and using ␤3 inserted into the Bluescript vector as a template. The identity of * This work was supported in part by Grants HL40387 and HL51258 from the National Institutes of Health. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18  Expression of Wild-type ␣IIb␤3 and ␣IIb␤3 Mutants in Human B Lymphocytes-pREP vectors containing cDNAs for wild-type and mutant ␣IIb and ␤3 were introduced into 7.5 ϫ 10 6 GM1500 B lymphocytes by electroporation (250 V and 960 millifarads). Stable cotransfectants were selected by growth in RPMI medium containing 20% fetal calf serum and both G418 (750 g/ml) and hygromycin (250 g/ml). The presence of ␣IIb␤3 on the lymphocyte surface was detected by flow cytometry. Cells were stained with the ␣IIb␤3-specific murine monoclonal antibody (mAb) A2A9 (9), the ␤3-specific mAb SSA6 (10), and the class-matched control antibody OKT3 (ATCC), followed by fluorosceinconjugated anti-murine IgG as described previously (11). Flow cytometry was performed using a FACScan flow cytometer (Becton-Dickinson). The assembly of ␣IIb␤3 heterodimers in lymphocytes was studied by pulse-chase analysis. Briefly, transfected lymphocytes were pulselabeled with [ 35 S]methionine (500 Ci/ml) for 60 min in methionine-free medium. The medium was then replaced with complete medium containing 1 mM unlabeled methionine, and the incubation was continued for various periods of time after which the cells were extracted with 2.5 mM Tris buffer, pH 7.4, containing 1% Triton X-100. Pro-␣IIb and ␣IIb␤3 were then immunoprecipitated from the extracts using polyclonal rabbit anti-␣IIb antiserum and the ␣IIb␤3-specific mAb A2A9, followed by SDS-polyacrylamide gel electrophoresis and autofluorography as described previously (12). The polyclonal anti-␣IIb antiserum used in these experiments only immunoprecipitates ␣IIb and pro-␣IIb and not ␣IIb␤3 heterodimers (data not shown). Integration of wild-type ␣IIb and the various ␣IIb mutants into the membranes of the lymphocyte endoplasmic reticulum (ER) was studied by treating isolated lymphocyte membranes with sodium carbonate, as described previously (13). Briefly, transfected lymphocytes were pulse-labeled with [ 35 S]methionine and chased with methionine-free medium for 4 h as described above. The labeled cells were then disrupted by 30 strokes in a Dounce homogenizer fitted with a tight pestle and the homogenate centrifuged for 10 min at 2000 rpm to sediment remaining intact cells and nuclei (14). Membranes were sedimented from the postnuclear supernatant by centrifugation at 230,000 ϫ g for 15 min and resuspended in 0.5-ml 100 mM sodium carbonate, pH 11.5. Following a 30-min incubation on ice, extracted proteins were separated from membranes by an additional centrifugation at 230,000 ϫ g for 2 h. After neutralization, 0.5% Triton X-100 was added to the samples and ␣IIb was immunoprecipitated using rabbit polyclonal anti-␣IIb antiserum. As a positive control for the sodium carbonate extraction, ␣IIb tr , a soluble ␣IIb mutant produced by introducing a stop codon into an ␣IIb cDNA proximal to nucleotides encoding the ␣IIb transmembrane domain, was expressed in COS-1 cells (15). Forty-eight hours after transfection, the COS-1 cells were carried through the sodium carbonate treatment protocol concurrently with the lymphocytes.
PMA-stimulated ␣IIb␤3 Function in Human B Lymphocytes-The ability of ␣IIb␤3 expressed in lymphocytes to interact with fibrinogen was tested by measuring the adherence of PMA (Sigma)-stimulated lymphocytes to immobilized fibrinogen (7) and the binding of soluble fluorescein isothiocyanate (FITC)-labeled fibrinogen to PMA-stimulated cells using flow cytometry (16). To measure ␣IIb␤3-mediated lymphocyte adherence to fibrinogen, the wells of Immulon 2 flat bottom microtiter plates (Dynatech Laboratories) were coated with 50 g/ml purified human fibrinogen in 50 mM NaHCO 3 buffer, pH 8.0, containing 150 mM NaCl. Unoccupied protein binding sites on the wells were blocked with 5 mg/ml bovine serum albumin dissolved in the same buffer. 1.5 ϫ 10 5 B lymphocytes, metabolically labeled overnight with [ 35 S]methionine, were suspended in 100 l of 50 mM Tris-HCl buffer, pH 7.4, containing 150 mM NaCl, 0.5 mM CaCl 2 , 0.1% glucose, and 1% bovine serum albumin and added to the protein-coated wells, either in the presence or absence of 200 ng/ml PMA. Following an incubation at 37°C for 30 min without agitation, the plates were vigorously washed four times with the lymphocyte suspension buffer and adherent cells were dissolved using 2% SDS. The SDS solutions were then counted for 35 S in a liquid scintillation counter. To compare the ligand binding activity of wild-type ␣IIb␤3 and the various ␣IIb␤3 mutants, adherence data were normalized to the level of heterodimer expression by dividing the percent adherence by the value for mean fluorescence intensity (MFI) obtained by flow cytometry. The normalization procedure was based on the high correlation between MFI and directly measured amounts of ␣IIb␤3 on the platelet surface (17). To measure the ␣IIb␤3mediated binding of soluble fibrinogen to lymphocytes, purified human fibrinogen (Kabi) (18) was labeled with FITC as described by Xia et al. (19). 1.5 ϫ 10 5 B lymphocytes were then suspended in 100 l of 10 mM sodium phosphate buffer, pH 7.4, containing 137 mM NaCl, 1 mM CaCl 2 , and 1% bovine serum albumin (suspension buffer) and incubated with 0.25 M FITC-labeled fibrinogen in the presence or absence of 200 ng/ml PMA for 30 min at 37°C. The cells were then washed once with the suspension buffer and resuspended in a fixation solution consisting of 10 mM sodium phosphate buffer, pH 7.4, containing 137 mM NaCl and 0.37% formalin. Following a 10-min incubation on ice, the cells were again washed once with the suspension buffer and analyzed by flow cytometry as described previously (16). In preliminary experiments to verify the method, washed platelets were substituted for lymphocytes. We detected no FITC-labeled fibrinogen binding to unstimulated platelets and there was a 3 log increase in MFI after platelet stimulation with PMA (data not shown).

Interaction of Lymphocytes Expressing ␣IIb␤3 with Soluble
Fibrinogen-Previously, we demonstrated that stimulation of EBV-transformed B lymphocytes expressing ␣IIb␤3 with the phorbol ester PMA results in the adherence of these cells to immobilized fibrinogen (7). However, it was uncertain whether this response to PMA results simply from more avid lymphocyte adherence to fibrinogen or from an actual increase in the affinity of ␣IIb␤3 for fibrinogen. To address this question, we tested the ability of soluble fibrinogen to inhibit lymphocyte adherence, based on the premise that soluble fibrinogen must bind to ␣IIb␤3 to inhibit adherence. Accordingly, we coated the wells of microtiter plates with four different concentrations of fibrinogen (1, 2, 5, and 10 g/ml) and measured PMA-stimulated lymphocyte adherence in the presence of increasing concentrations of soluble fibrinogen. As shown in Fig. 2A, the adherence of PMA-stimulated lymphocytes expressing ␣IIb␤3 decreased as the concentration of soluble fibrinogen increased for each concentration of immobilized fibrinogen, suggesting that there was competition between the soluble and immobilized fibrinogen for binding to ␣IIb␤3. At the highest concentration of soluble fibrinogen tested (14 mg/ml or 44 M), lymphocyte adherence was inhibited by Ϸ90% to plates that had been coated with fibrinogen at 5 and 10 g/ml and by Ϸ95% to plates that had been coated at 1 and 2 g/ml. To verify that the inhibitory effect of soluble fibrinogen was specific, we tested the ability of equimolar concentrations of a number of other soluble proteins to inhibit lymphocyte adherence. As shown in Fig. 2B, other proteins including the ␤3-specific mAb SSA6, the ␣IIbspecific mAb B1B5, albumin, and transferrin failed to inhibit PMA-stimulated lymphocyte adherence at the maximum concentration of soluble fibrinogen examined, 44 M.
The ability of soluble fibrinogen to inhibit PMA-stimulated lymphocyte adherence to immobilized fibrinogen is indirect evidence that soluble fibrinogen can bind to ␣IIb␤3 on these cells. Therefore, to directly demonstrate soluble fibrinogen binding to lymphocyte ␣IIb␤3, we labeled fibrinogen with FITC and measured FITC-labeled fibrinogen binding to untransfected lymphocytes and lymphocytes expressing ␣IIb␤3 by fluorescence-activated flow cytometry (16). As shown in Fig. 3, there was no change in the position of the fluorescence histogram when untransfected lymphocytes incubated with FITC- labeled fibrinogen were stimulated with PMA, demonstrating that there were no detectable binding sites for FITC-labeled fibrinogen on the surface of these cells. Additionally, the positions of these histograms coincided with the position of the fluorescence histogram of lymphocytes expressing ␣IIb␤3 that had been incubated with FITC-labeled fibrinogen in the absence of PMA stimulation. However, following PMA stimula-tion, the position of the latter histogram was shifted to the right, indicating that PMA had exposed fibrinogen binding sites on the lymphocyte surface. To verify that these binding sites corresponded to ␣IIb␤3, the fibrinogen binding measurements were repeated in the presence of A2A9, a mAb that specifically inhibits fibrinogen binding to ␣IIb␤3 on stimulated platelets (9). The presence of A2A9 shifted the position of the histogram back to that of unstimulated cells, confirming that the FITC-labeled fibrinogen was bound to ␣IIb␤3. Chelation of divalent cations using EDTA gave an identical result, verifying that the measured fibrinogen binding was cation-dependent (data not shown). Thus, these experiments indicate that PMA stimulation results in an increase in the affinity of lymphocyte ␣IIb␤3 for soluble, as well as immobilized, fibrinogen. Moreover, they argue that the adherence of lymphocytes expressing ␣IIb␤3 to immobilized fibrinogen is a reflection of this increased affinity rather than a post-receptor event.
Effect of Mutations Involving the ␣IIb Cytoplasmic Tail on the Expression of ␣IIb␤3 by Lymphocytes-To examine the role of the ␣IIb cytoplasmic tail in agonist-stimulated ␣IIb␤3 function, we truncated ␣IIb proximal and distal to its GFFKR motif by replacing the codons for Gly-991 and Asn-996 in an ␣IIb cDNA with stop codons. The resulting ␣IIb mutants were designated ␣IIb 990 and ␣IIb 995 after their carboxyl-terminal residue (Fig. 1). In addition, we replaced GFFKR with AAAAA (␣IIb A 5 ) and Gly-991 with the helix-breaking amino acid Pro (␣IIb G991P ) based on predictions derived from homology modeling that the membrane-proximal region of the ␣IIb tail including GFFKR assumes an ␣ helical configuration (6). 2 The resulting ␣IIb mutants and wild-type ␣IIb were then coexpressed with ␤3 in EBV-transformed B lymphocytes. To assess the level of expression of each ␣IIb␤3 mutant on the lymphocyte surface, the cell lines were stained with the ␣IIb␤3-specific mAb A2A9 and examined by flow cytometry. As shown in Fig.  4, there was similar staining of wild-type ␣IIb␤3, ␣IIb 995 ␤3, and ␣IIb G991P ␤3, whereas the staining of lymphocytes expressing ␣IIb 990 ␤3 and ␣IIb A 5 ␤3 was markedly decreased. In six separate transfections of ␣IIb 990 and three separate transfections of ␣IIb A 5 , there was only a 2.6 Ϯ 0.5-fold and a 1.9 Ϯ 0.3-fold increase in MFI, respectively, when the cells were 2 T. Keiber-Emmons and J. S. Bennett, unpublished data.

FIG. 2. Inhibition of lymphocyte adherence to immobilized fibrinogen by soluble fibrinogen.
A, the wells of microtiter platelets were coated with purified human fibrinogen at concentrations of 10 g/ml (f), 5 g/ml (Ⅺ), 2 g/ml (q), and 1 g/ml (E). Lymphocytes expressing ␣IIb␤3 were metabolically labeled with [ 35 S]methionine and incubated with the immobilized fibrinogen at 37°C in the presence of 200 ng/ml PMA and increasing concentrations of soluble fibrinogen. Lymphocyte adherence to the immobilized fibrinogen was then quantitated after 30 min as described under "Experimental Procedures." B, to demonstrate the specificity of the inhibitory effect of soluble fibrinogen seen in A, the indicated proteins at concentrations of 44 M were substituted for soluble fibrinogen in the adherence assay. SSA6 is a murine mAb specific for ␤3; BIB5 is a murine mAb specific for ␣IIb. Control adherence was PMA-stimulated lymphocyte adherence in the absence of soluble inhibitor and was designated 100%. The wells of the microtiter plates used for these experiments were coated with fibrinogen at a concentration of 5 g/ml.
FIG. 3. Flow cytometry histograms resulting from soluble fibrinogen binding to unstimulated and PMA-stimulated control GM1500 lymphocytes and to GM1500 lymphocytes expressing ␣IIb␤3. Purified human fibrinogen was labeled with fluorescein isothiocyanate as described under "Experimental Procedures." Control GM1500 lymphocytes and GM1500 lymphocytes expressing ␣IIb␤3, suspended at a concentration of 1.5 ϫ 10 5 cells/100 l in 10 mM sodium phosphate buffer, pH 7.4, containing 137 mM NaCl, 0.5 mM CaCl 2 , and 1% bovine serum albumin were incubated with 0.25 M fluoresceinlabeled fibrinogen (FITC-Fib.) in the absence or presence of 200 ng/ml PMA at 37°C. After 30 min, the cells were washed with suspension buffer, fixed in buffer containing 0.37% formalin, and examined by flow cytometry. The specificity of fibrinogen binding was determined by performing the incubation in the presence of the inhibitory mAb A2A9 at a concentration of 50 g/ml.
␣IIb␤3 Cytoplasmic Domain Mutations stained with A2A9 compared with staining with the classmatched control antibody OKT3. In contrast, the increases in MFI after staining cells expressing wild-type ␣IIb␤3, ␣IIb 995 ␤3, and ␣IIb G991P ␤3 were 15.3 Ϯ 2-fold, 5.9 Ϯ 0.7-fold, and 9.7 Ϯ 2.8-fold, respectively. Staining the cells with the ␤3-specific mAb SSA6 instead of A2A9 gave similar results, indicating that the reduction in A2A9 staining was not due to absence of the epitope for A2A9 on cells expressing ␣IIb 990 ␤3 and ␣IIb A 5 ␤3 (data not shown). Thus, these data indicate that the GFFKR motif plays an important role in ␣IIb␤3 expression, a conclusion that is consistent with previous observations of the role played by GFFKR in the expression of ␤1 and ␤2 integrins (20 -22).
To examine the basis for the reduced expression of ␣IIb 990 ␤3 and ␣IIb A 5 ␤3 on the lymphocyte surface, we performed pulsechase studies, comparing the biosynthesis of these heterodimers with that of wild-type ␣IIb␤3 and ␣IIb 995 ␤3. Lymphocytes were pulsed with [ 35 S]methionine for 1 h and at intervals throughout the subsequent 8-h chase, aliquots of lymphocytes were extracted with Triton X-100 and pro-␣IIb and ␣IIb␤3 heterodimers were immunoprecipitated. As seen in Fig.  5, there was no discernible difference in either the synthesis or stability of pro-␣IIb, pro-␣IIb 995 , pro-␣IIb 990 , or pro-␣IIb A 5 throughout the 8-h period of chase. However, whereas heterodimers were clearly present by 2 h in detergent extracts of the ␣IIb␤3 and ␣IIb 995 ␤3 transfectants, only traces of heterodimer were detectable in extracts from the ␣IIb 990 ␤3 and ␣IIb A 5 ␤3 transfectants at any time point up to 8 h. Thus, these studies indicate that the reduced expression of ␣IIb 990 ␤3 and ␣IIb A 5 ␤3 resulted from decreased assembly of ␣IIb␤3 heterodimers and suggest that deletion or replacement of GFFKR impairs the ability of ␣IIb to interact with ␤3.
Assembly of ␣IIb and ␤3 into heterodimers involves sequences located in the extracellular domain of each subunit (15,23), whereas GFFKR is located in the ␣IIb cytoplasmic tail adjacent to its transmembrane sequence. Thus, it is unlikely that absence of GFFKR directly impairs the ability of ␣IIb to associate with ␤3. On the other hand, it is conceivable that the absence of GFFKR impairs the stop-transfer mechanism that anchors ␣IIb in membranes, thereby shifting a fraction of nascent ␣IIb from the membrane to the lumen of the ER, where it would be unavailable to associate with nascent ␤3. Alternatively, absence of GFFKR could affect the folding of nascent ␣IIb such that it was no longer able to be recognized by ␤3. To address the first possibility, we studied the association of ␣IIb, ␣IIb 995 , ␣IIb 990 , and ␣IIb A 5 with lymphocyte membranes by isolating membranes from [ 35 S]methionine-labeled transfected cells and treating the membranes with 0.1 M sodium carbonate, pH 11.3 (13). This treatment converts closed membrane vesicles into open sheets, thereby releasing their lumenal contents and peripheral membrane proteins in soluble form. As concurrent positive and negative controls, membranes were also prepared from [ 35 S]methionine-labeled COS-1 cells transiently expressing ␣IIb tr , an ␣IIb mutant truncated proximal to the ␣IIb transmembrane domain (15), and from [ 35 S]methionine-labeled untransfected lymphocytes. As shown in Fig. 6, soluble ␣IIb tr was present in the sodium carbonate extract of COS-1 cell membranes, but neither ␣IIb, ␣IIb 995 , ␣IIb 990 , nor ␣IIb A 5 were detected in the sodium carbonate extracts of lymphocyte membranes. Thus, these experiments indicate that absence of GFFKR does not prevent ␣IIb anchoring in lymphocyte membranes, implying that each form of nascent ␣IIb is available to associate with nascent ␤3 in the lymphocyte ER. Consequently, the failure to detect ␣IIb␤3 in lymphocytes expressing ␣IIb 990 and ␣IIb A 5 suggests that absence of GFFKR results in a conformational change in ␣IIb that impairs its ability to be recognized by ␤3.
Effect of Mutations Involving the ␣IIb Cytoplasmic Tail on the Adherence of Lymphocytes to Immobilized Fibrinogen-Next, we compared the ability of mock-transfected lymphocytes and lymphocytes expressing ␣IIb␤3 heterodimers containing wild-type ␣IIb, ␣IIb 990 , or ␣IIb 995 to adhere to immobilized fibrinogen. As shown in Fig. 7, there was no adherence of mock-transfected lymphocytes to fibrinogen, either before or after PMA stimulation. By contrast, we found that 4 -8% of lymphocytes expressing wild-type ␣IIb␤3 were adherent to fibrinogen in the absence of PMA stimulation and adherence increased to 35-40% following stimulation by 200 ng/ml PMA. Moreover, deleting the 13 amino acids in the ␣IIb tail distal to ␣IIb␤3 Cytoplasmic Domain Mutations GFFKR had no effect on lymphocyte adherence because 12-20% of lymphocytes expressing ␣IIb 995 ␤3 were adherent in the absence of PMA stimulation and stimulation increased adherence to 35-50%. In addition, although only 1-2% and 8 -9% of lymphocytes expressing ␣IIb 990 ␤3 were adherent in the absence and presence of PMA stimulation, respectively, when lymphocyte adherence was normalized for the level of heterodimer expression by calculating a ratio of adherence (Fig. 7) to MFI (Fig. 4), the ratios for cells expressing ␣IIb 990 ␤3 (0.25) and ␣IIb␤3 (0.17) were similar. The adherence of lymphocytes expressing wild-type ␣IIb␤3 and each of the ␣IIb mutants was inhibited completely by the mAb A2A9, confirming that it was mediated by the extracellular domain of ␣IIb␤3 (data not shown). Thus, these data indicate that lymphocytes expressing ␣IIb 990 ␤3 adhere to fibrinogen at least as avidly as cells expressing wild-type ␣IIb␤3. In addition, they demonstrate that truncation of the ␣IIb cytoplasmic tail, whether proximal or distal to GFFKR, neither impairs ␣IIb␤3 function in lymphocytes nor precludes the ability of ␣IIb␤3 to respond to PMA stimulation in these cells.
The results described above suggest that the presence or absence of GFFKR in the cytoplasmic tail of ␣IIb may affect the conformation of ␣IIb and its ability to interact with ␤3. To pursue this observation, we measured the unstimulated and PMA-stimulated adherence of lymphocytes expressing ␣IIb A 5 ␤3 and ␣IIb G991P ␤3 to fibrinogen. Like ␣IIb 990 ␤3, we found the ability of ␣IIb A 5 ␤3 to mediate lymphocyte adherence was commensurate with its level of expression on the lymphocyte surface, since 4 -5% of the cells were adherent in the FIG. 5. Pulse-chase analysis of ␣IIb␤3 biosynthesis in lymphocytes. Lymphocytes expressing wild-type ␣IIb␤3 and the ␣IIb 990 ␤3, ␣IIb 995 ␤3 and ␣IIb A 5 ␤3 mutants were pulse-labeled with [ 35 S]methionine for 1 h and chased with medium containing 1 mM unlabeled methionine for 8 h. Aliquots of the cells were extracted at the indicated times during the chase with buffer containing 1% Triton X-100. Pro-␣IIb was immunoprecipitated from the extracts using polyclonal ␣IIb antisera, and ␣IIb␤3 was immunoprecipitated using the ␣IIb␤3-specific mAb A2A9. Pro-␣IIb is the single chain ␣IIb precursor, and ␣IIb␣ is the ␣IIb heavy chain.  ␣IIb␤3 Cytoplasmic Domain Mutations absence of PMA stimulation and adherence increased to 13-14% following exposure to PMA (Fig. 7). It is noteworthy, however, that while the adherence of PMA-stimulated lymphocytes expressing ␣IIb G991P ␤3 (38 -41%) was identical to that of cells expressing wild-type ␣IIb␤3, the adherence of unstimulated cells (17-21%) was significantly increased (p Ͻ 10 Ϫ7 ). This suggests that replacing Gly-991 with proline resulted in an increase in the fraction of ␣IIb␤3 heterodimers that reside on the lymphocyte surface in a constitutively active state.
Effect of Mutations Involving the ␤3 Cytoplasmic Tail on ␣IIb␤3 Function-Our previous studies of the function of ␣IIb␤3-␣L␤2 chimeras suggested that the ␤3 cytoplasmic tail is specifically involved in ␣IIb␤3 activation in lymphocytes (7). To test this suggestion, we introduced two naturally occurring mutations known to impair ␣IIb␤3 function, Ser-752 3 Pro (4) and Arg-724 3 stop (3), into the cytoplasmic tail of ␤3 by in vitro mutagenesis (Fig. 1) and measured the adherence of lymphocytes expressing ␣IIb␤3 heterodimers containing these mutations to immobilized fibrinogen. As shown by the flow cytometry histograms of cells stained with the mAb A2A9 in Fig. 8, comparable levels of wild-type ␣IIb␤3, ␣IIb␤3 containing the Ser-752 3 Pro mutation (␣IIb␤3 S752P ), and ␣IIb␤3 containing ␤3 truncated at amino acid 724 (␣IIb␤3 724 ) were expressed on the surface of transfected lymphocytes. Nevertheless, whereas 4.8% of lymphocytes expressing wild-type ␣IIb␤3 were adherent to fibrinogen in the absence of PMA stimulation and adherence increased to 37.8% in the presence of PMA, the corresponding values for lymphocytes expressing ␣IIb␤3 S752P and ␣IIb␤3 724 were 0.7% and 11.7% and 0.7% and 15.2%, respectively (Fig. 9). Thus, these experiments confirm the importance of the ␤3 cytoplasmic tail in agonist-stimulated ␣IIb␤3 function.
Despite levels of expression comparable with wild-type ␣IIb␤3, ␣IIb␤3 S752P and ␣IIb␤3 724 were only 25-30% as effective in mediating lymphocyte adherence to fibrinogen, indicating that their function was impaired. On the other hand, each mutant is unable to support platelet aggregation and fibrinogen binding (4) and when expressed in CHO cells, they greatly reduce ligand binding by constitutively active ␣IIb␤3 (6). Accordingly, we had anticipated that the adherence of lymphocytes expressing ␣IIb␤3 S752P and ␣IIb␤3 724 would be essentially absent. Previously, we noted that B lymphocytes express a small amount of endogenous ␤3 as the ␤ subunit of ␣v␤3 (7). Thus, it is possible that when we transfected lymphocytes with ␣IIb and ␤3 S752P or ␤3 724 , the resulting heterodimers were a mixture of wild-type and mutant molecules. Consequently, wild-type ␣IIb␤3 could have responsible for the reduced level of lymphocyte adherence we observed. To examine this possibility, we transfected lymphocytes with cDNAs for both ␣IIb and ␤3 or with a cDNA for ␣IIb alone. As seen in Fig. 10A, transfection with ␣IIb alone resulted in the substantial accumulation of ␣IIb␤3 on the lymphocyte surface (MFI of 98 versus 167 for the ␣IIb␤3 transfectant). Moreover, as seen in Fig. 10B, the ␣IIb␤3 expressed by these cells mediated adherence to fibrinogen at a level that was commensurate with the amount of ␣IIb␤3 expressed on their surface. Thus, it is plausible that the reduced adherence of cells expressing ␣IIb␤3 S752P and ␣IIb␤3 724 was actually mediated by coincidentally expressed wild-type ␣IIb␤3.

DISCUSSION
In platelets, agonist-initiated signals convert ␣IIb␤3 from an inactive to an active state by increasing its affinity for ligands such as fibrinogen and von Willebrand factor (1). We studied this process in vitro using a B lymphocyte expression system based on the premise that agonist-initiated inside-out signaling in lymphocytes can up-regulate the function of the heterologously expressed ␣IIb␤3 much as it up-regulates the function of endogenous lymphocyte integrins (24). We found that PMA-stimulated, but not unstimulated, lymphocytes expressing ␣IIb␤3 were able to adhere to immobilized fibrinogen (7). Nevertheless, the question remained whether the PMA-stimulated lymphocyte adherence we observed simply resulted from more avid adherence of lymphocytes to fibrinogen, a post-receptor mechanism perhaps due to a cytoskeletal rearrangement (25,26), or resulted from an actual increase in the affinity of individual ␣IIb␤3 heterodimers for fibrinogen, analogous to the increase in ␣IIb␤3 affinity that occurs when platelets are stimulated by agonists (18). The increased affinity of ␣IIb␤3 in platelets is manifest as the ability of ␣IIb␤3 to bind soluble fibrinogen (18). Accordingly, we tested the ability of the ␣IIb␤3 expressed by transfected lymphocytes to bind soluble fibrinogen. First, we found that soluble fibrinogen specifically inhibited lymphocyte adherence to immobilized fibrinogen, implying that PMA stimulation provided ␣IIb␤3 access to both soluble and immobilized ligand. Second, we detected the binding of soluble fibrinogen directly to ␣IIb␤3 on PMA-stimulated lymphocytes using flow cytometry. Although the shift in fluorescence following PMA stimulation of lymphocytes in these experiments was substantially less than that of platelets, measurement of the number of ␣IIb␤3 molecules on transfected lymphocytes using 125 I-labeled A2A9 and comparison of the fluorescence histograms of lymphocytes and platelets stained with A2A9 indicate that there are only 15-20% as many ␣IIb␤3 molecules on transfected lymphocytes as there are on plate-lets. 3 Additionally, studies of PMA-stimulated soluble ICAM-1 binding to ␣L␤2 on T lymphocytes suggest that PMA stimulation converts an average of only 16% of the ␣L␤2 to a high affinity state (29). Similarly, it has been observed that a mAb that binds exclusively to the activated conformation of ␣M␤2 binds only to 10 -20% of the ␣M␤2 molecules on activated neutrophils (30). Thus, it is possible that, unlike platelets, where there is a 1:1 correlation between fibrinogen binding and the number of ␣IIb␤3 molecules (9), only a fraction of the ␣IIb␤3 molecules on transfected B lymphocytes bind fibrinogen following PMA stimulation. Nevertheless, our data clearly demonstrate that, like platelets, agonist stimulation of lymphocytes increases the affinity of ␣IIb␤3 for fibrinogen and provide evidence that this increase in affinity is responsible for the adherence of the cells to immobilized fibrinogen.
The mechanism by which inside-out signaling increases the affinity of ␣IIb␤3 and other integrins for ligands is not known. Although the responsible intracellular signals interact with the cytoplasmic portions of integrins, the relative contribution of each integrin subunit to the signaling process is uncertain. Thus, while there is convincing evidence for the participation of the ␤ subunit tail in the response of many integrins to agonists (29), the contribution of the ␣ subunit tail appears to depend on the integrin studied and the cell in which the integrin is expressed. For example, truncation of the cytoplasmic tails of ␣2, ␣4, and ␣6A immediately distal to their GFFKR motifs abolished the constitutive and phorbol ester-stimulated ligand binding activity of ␤1 integrins in RD (20), K562 (21), and P338D 1 (30) cells, whereas replacement of the ␣2 tail with tails of ␣4 or ␣5 (20) and the ␣4 tail with those of ␣2 or ␣5 had no effect (21). Thus, these data indicate that an ␣ subunit tail can be required for integrin function, but this requirement may be fulfilled by the tails of a number of ␣ subunits. Further, the addition of only 3-4 amino acids on the carboxyl side of GFFKR may be sufficient to restore the ligand binding activity of integrins containing truncated ␣ subunits (31). In contrast to these observations, deletion of the ␣L tail distal to GFFKR had no effect on the ability of ␣L␤2 expressed in COS cells to interact with ICAM-1 (22), nor did the deletion of the ␣M tail impair the interaction of ␣M␤2 with iC3b-coated erythrocytes (32).
␣IIb␤3 function has been studied extensively in vitro using adherent CHO fibroblasts (5,6). Although CHO cells expressing ␣IIb␤3 spontaneously adhere to and spread on immobilized fibrinogen (33), they are unable to bind the activation-dependent, ␣IIb␤3-specific mAb PAC1 (34), suggesting that their ␣IIb␤3 is in a low affinity state. However, truncation of either the ␣IIb tail before GFFKR or the ␤3 tail before the membraneproximal sequence LLITIHD results in PAC1 binding, implying that each truncation shifts ␣IIb␤3 from a low affinity to a high affinity state (5,6). These experiments suggest a model of ␣IIb␤3 activation in which GFFKR and LLITIHD act as a "hinge," interacting in unstimulated cells to maintain ␣IIb␤3 in a low affinity state and separating following agonist stimulation to propagate signals from the cytoplasmic to the extracellular portions of the molecule (6). In support of the model, disruption of either the GFFKR or LLITIHD sequence by alanine substitutions resulted in PAC1 binding, as did ␣IIb R995D and ␤3 D723R mutations (35). Moreover, the "reversal" mutant ␣IIb R995D ␤3 D723R was inactive, arguing that a salt bridge between Arg-995 in ␣IIb and Asp-723 in ␤3 in unstimulated cells constrains ␣IIb␤3 function. Nevertheless, replacing the ␣IIb cytoplasmic tail with the tails of ␣2, ␣5, ␣6A, or ␣6B also induces PAC1 binding, despite preservation of the GFFKR FIG. 10. ␣IIb␤3 expression on the surface of lymphocytes transfected both ␣IIb and ␤3 or with ␣IIb alone. A, histograms from the flow cytometry of mock-transfected cells, cells transfected with ␣IIb and ␤3, and cells transfected with ␣IIb alone. Cells were stained with the ␣IIb␤3-specific mAb A2A9 and analyzed as described in Fig. 2 and under "Experimental Procedures." B, comparison of the unstimulated and PMA-stimulated adherence of the transfected lymphocytes to fibrinogen. Lymphocyte adherence was measured as described in Fig. 7 and under "Experimental Procedures." motif (6). Thus, other features of the ␣IIb tail in CHO cells, besides the presence of GFFKR, may regulate ␣IIb␤3 function.
We have examined the contribution of the ␣IIb and ␤3 cytoplasmic tails to ␣IIb␤3 function by expressing ␣IIb␤3 in the B lymphocytes (7). We found that despite deletion of the ␣IIb cytoplasmic tail, ␣IIb␤3 retained the ability to mediate lymphocyte adherence to fibrinogen, regardless of whether the tail was deleted before or after the GFFKR motif. Thus, like ␣L␤2 (22) and ␣M␤2 (32), the presence of an ␣IIb tail is not required for ␣IIb␤3 function in these cells. Moreover, we found that ␣IIb␤3 retained the ability to be up-regulated by PMA despite the absence of a GFFKR motif. Thus, in contrast to ␣IIb␤3 in CHO cells (5), our data indicate that the bulk of the ␣IIb␤3 in unstimulated lymphocytes remains in an inactive state despite the absence of GFFKR and suggest it unlikely that GFFKR functions as part of a hinge in lymphocytes, although it possibly could do so in other cells. Nevertheless, the proportion of lymphocytes expressing ␣IIb G991P ␤3 that were adherent to fibrinogen in the absence of PMA stimulation was significantly greater than the proportion of unstimulated lymphocytes expressing wild-type ␣IIb␤3 (p Ͻ 10 Ϫ6 ). Thus, mutations involving GFFKR in lymphocytes, as in CHO cells, appear to increase the fraction of lymphocytes expressing ␣IIb␤3 in an activated form.
We found that the major consequence of removing the GFFKR motif from ␣IIb was a marked reduction in ␣IIb␤3 expression on the lymphocyte surface. A similar effect of GFFKR deletion on integrin expression has been observed when ␣2, ␣4, ␣5, and ␣L were truncated proximal to GFFKR (20,21,30,36) and when ␣IIb, truncated proximal to GFFKR, was expressed in transient expression systems (6). Although a number of explanations for the impaired ␣IIb␤3 expression are possible, including a truncation-related decrease in ␣IIb synthesis and/or stability or the intracellular retention of misfolded or activated ␣IIb␤3 heterodimers, we found that removing GFFKR impaired the ability of ␣IIb to associate with ␤3. This was unexpected because ␣IIb␤3 assembly involves sequences located in the ␣IIb and ␤3 extracellular domains at considerable distances from their cytoplasmic tails (15,23). However, GFFKR is located adjacent to the ␣IIb transmembrane domain. Thus, it is conceivable that its absence impairs the stop-transfer mechanism that anchors ␣IIb in membranes (37), thereby shifting a fraction of nascent ␣IIb from the membrane to the lumen of the ER, where it might be unavailable to associate with nascent ␤3. We addressed this possibility by treating isolated lymphocyte membranes with 0.1 M sodium carbonate, pH 11.5, to extract proteins confined to the ER lumen. In control experiments, we found that ␣IIb truncated proximal to its transmembrane domain was extracted from isolated COS cell membranes by this treatment, and in agreement with previous studies using isolated dog pancreas microsomes (38), we found that intact ␣IIb was not extractable from isolated lymphocyte membranes. We were never able to detect even traces of ␣IIb 995 , ␣IIb 990 , or ␣IIb A 5 in sodium carbonate extracts. Thus, these results indicate that both ␣IIb 990 and ␣IIb A 5 were integrated into lymphocyte membranes, eliminating the possibility that removal of GFFKR shifts ␣IIb from the ER membrane to the ER lumen.
A second possible explanation for our results is that the absence of GFFKR induces a change in the conformation of ␣IIb that is propagated to its extracellular domain, thereby impairing its recognition by ␤3. The inability of membrane-associated ␣IIb lacking GFFKR to recognize ␤3 provides prima facie evidence that such an event occurs when GFFKR is either deleted or mutated. Moreover, propagation of conformational change across the length of the molecule is a property of integrins that accounts for both inside-out and outside-in signaling (2,39,40). In addition, we observed that when ␣IIb 990 , lacking its 18 carboxyl-terminal amino acids, was extracted from lymphocyte membranes using sodium carbonate, it migrated as doublet on SDS gels whose mobility was retarded compared with the mobility of wild-type ␣IIb (Fig. 6). The migration of the related ␣ subunit, ␣v, truncated distal to GFFKR at residue 995, was also found to be retarded on SDS gels (41). This was thought to be due to altered ␣v folding because of an accompanying change in the pattern of ␣v cleavage by chymotrypsin. Thus, it is possible that the retarded migration of ␣IIb 990 also resulted from a truncation-induced alteration in ␣IIb folding. Loss of GFFKR, however, cannot be the entire explanation because the electrophoretic mobility of wild-type ␣IIb and ␣IIb A 5 were identical. Nevertheless, the absolute conservation of GFFKR, and the related GFFRR sequence (42), among integrin ␣ subunits implies that this motif supplies an indispensable function. Our data suggest that GFFKR is required for proper ␣ subunit folding. Perhaps the shift toward increased ␣IIb␤3 activity observed when GFFKR is replaced or mutated reflects an impairment of this requirement.
Our finding that ␣IIb␤3 can be activated and adhere to fibrinogen, despite the absence of an ␣IIb tail, suggests that the ␤3 tail is primarily involved in agonist-mediated regulation of ␣IIb␤3 function, at least in lymphocytes. The diminished adherence of lymphocytes expressing ␣IIb␤3 S752P and ␣IIb␤3 724 , despite the presence of intact ␣IIb, is consistent with this conclusion, as are previous studies of ␣L␤2 function (22,43). Accordingly, our data are consistent with a model of ␣IIb␤3 activation in which signals interacting with the cytoplasmic tail of ␤3 are propagated to the extracellular domain of the heterodimer to expose its ligand binding site. However, differences in ␣IIb␤3 function in various cells, for example lymphocytes and CHO cells, also suggest that host cell factors interacting with ␣IIb␤3 may differ. Because lymphocytes are of hematopoietic origin and normally express ␤3, the factors in lymphoocytes may be the same or related to those present in platelets. The recent identification of a protein from T cells, cytohesin-1, related to the yeast SEC7 gene product, that specifically interacts with the ␤2 cytoplasmic tail to induce ␤2 integrin-dependent adherence of Jurkat cells to ICAM-1 (44), provides support for our model and suggests that the same or a similar protein could interact with the ␤3 cytoplasmic tail to activate ␣IIb␤3. A polypeptide called ␤3-endonexin that interacts specifically with the ␤3 cytoplasmic tail has been isolated from a B cell cDNA library (45). Whether this polypeptide functions in manner analogous to cytohesin-1 is not yet known.